Determining Seed Vigor with Hyperspectral Imaging

University of Florida

Lucia Carrero

Professors: Dr. Rowland and Dr. Zare

Co-Mentors: Justin Pitts and Weihuang Xu

1

This material is based upon work supported by the IoT4Ag Engineering Research Center funded by the National Science Foundation (NSF) under NSF Cooperative Agreement Number EEC-1941529.  Any opinions, findings and conclusions, or recommendations expressed in this material are those of the author(s), and do not necessarily reflect those of the NSF.

IoT4Ag

JULY 28, 2021

REU PRESENTATION

Introduction

2

Seed Vigor Classification allows for determining which seeds will be more successful.

IoT4Ag

JULY 28, 2021

REU PRESENTATION

Overview of the Technical Approach

• Hyperspectral images are taken to predict seed vigor.

​

• These seeds are then grown on sheets. The root length and thickness are measured.

​

• Data is statistically compared.

3

IoT4Ag

JULY 28, 2021

REU PRESENTATION

Hyperspectral Imaging

• Images that contain the information of the presence of different light wavelengths for each pixel.

4

Gatorsense Hyperspectral Analysis Introduction

IoT4Ag

JULY 28, 2021

REU PRESENTATION

Image Processing

• The steps for usable HSI data are converting the image to RGB, Bootstrapping, K-Means segmentation.

5

RGB

Bootstrapping

K-means

IoT4Ag

JULY 28, 2021

REU PRESENTATION

K-Means

• Used to group the data we are interested in.

​

• In our case, it will differentiate the peanuts from the background.

​

6

Towards Data Science

​

IoT4Ag

JULY 28, 2021

REU PRESENTATION

Sample Image

7

IoT4Ag

JULY 28, 2021

REU PRESENTATION

Bootstrap Sampling Method

8

Towards Data Science

​

IoT4Ag

JULY 28, 2021

REU PRESENTATION

Plot for Average Spectrum – Cultivar FLO331 with pod, from top

9

IoT4Ag

JULY 28, 2021

REU PRESENTATION

Harvesting Date

10

IoT4Ag

JULY 28, 2021

REU PRESENTATION

11

IoT4Ag

JULY 28, 2021

REU PRESENTATION

Results

12

IoT4Ag

JULY 28, 2021

REU PRESENTATION

Next Steps

• Further data analysis to better understand the relationship of seed vigor and hyperspectral information.

13

IoT4Ag

JULY 28, 2021

REU PRESENTATION

Acknowledgements and Reflection

• Thank you!
• Dr. Zare and Dr. Rowland
• Co-mentors Weihuang Xu and Justin Pitts
• IoT4Ag and both the Machine Learning and Sensing Laboratory and the Center for Stress Resilient Agriculture

14

IoT4Ag

JULY 28, 2021

REU PRESENTATION

References

• Gatorsense Hyperspectral Analysis Introduction

https://github.com/GatorSense/HyperspectralAnalysisIntroduction

​

• Towards Data Science

https://towardsdatascience.com/an-introduction-to-the-bootstrap-method-58bcb51b4d60

​

https://towardsdatascience.com/understanding-k-means-clustering-in-machine-learning-6a6e67336aa1

​

​

15

IoT4Ag

JULY 28, 2021

REU PRESENTATION